R/zeroInd.R

In netgsa: Network-Based Gene Set Analysis

zeroInd <-
function(Amat, r){
    if (sum(t(Amat)!=Amat)>0){
      stop("This method only works for symmetric matrix!")
    }
    p <- dim(Amat)[1]
    oneMat <- matrix(0, p, p)
    zeroMat <- matrix(0, p, p)
    
    one.pos <- which(Amat!=0, arr.ind = TRUE)
    zero.pos <- which(Amat==0, arr.ind = TRUE)

    zero.pos <- zero.pos[which(zero.pos[,1] > zero.pos[,2]) ,]
    if (length(zero.pos)==0){
      warning("The matrix is zero!")
      zeroMat <- matrix(0, p, p)
      zeroArr <- NULL
    } else{
      if (is.null(dim(zero.pos))){
        zero.pos <- matrix(zero.pos, nrow = 1)
      }
      sel.zero <- sample(seq(1, nrow(zero.pos)), r * nrow(zero.pos), replace = FALSE) 
      zeroMat[zero.pos[sel.zero, ]] <- 1
      zeroMat <- zeroMat + t(zeroMat)  
      zeroArr <- zero.pos[sel.zero, ]
      
    }
    out <- list()
    out$zeroArr = zeroArr
    out$zeroMat = zeroMat
    
    if (dim(one.pos)[1] == 0){
      warning("The matrix is zero!")
      out$oneMat = matrix(0, p, p)
    } else 
    {
      one.pos <- one.pos[which(one.pos[,1] > one.pos[,2]) ,]
      if (is.null(dim(one.pos))){
        one.pos = matrix(one.pos, nrow = 1)
      }
      sel.one <- sample(seq(1, nrow(one.pos)), r * nrow(one.pos), replace = FALSE) 
      oneMat[one.pos[sel.one, ]] <- 1
      oneMat <- oneMat + t(oneMat)
      diag(oneMat) <- 0
      
      out$oneMat = oneMat  
    }
    
    return(out)  
  }